# US 2025/0232843 A2 · Systems and methods for plant process optimisation

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0232843 A2
- Document ID: US20250232843A2
- Status: Published application
- Kind: A2
- Application number: 18/273,371
- Filed: 2021-01-20
- Published: 2025-07-17
- Earliest claimed: 2021-01-20
- Assignees: Jems, Energetska Druzba, D.O.O.
- Applicants: Jems, Energetska Druzba, D.O.O.
- Inventors: Johannes J. KRMC; Mitja JERMOL; Ervin PFEIFER; Suzana LEBEN
- Web page: https://patentgrep.com/patent/US20250232843A2

## CPC

- G05B 13/0265: {the criterion being a learning criterion}
- G16C 20/10: Analysis or design of chemical reactions, syntheses or processes
- G05B 13/048: {using a predictor}
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- G16C: COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE

## Abstract

A control system for optimising a fuel generation process performed by a synthetic fuel generation plant. The system comprises an interface unit configured to be communicably coupled to the synthetic fuel generation plant; a digital twin representative of the synthetic fuel generation plant; and an optimisation unit in communication with the interface unit and the digital twin The optimisation unit is configured to perform an optimisation process to determine, from an initial state of the synthetic fuel generation plant, an updated state of the fuel generation plant using the digital twin.

## Claims

### Claim 1

A control system for optimising a fuel generation process performed by a synthetic fuel generation plant, the control system comprising:
an interface unit configured to be communicably coupled to the synthetic fuel generation plant, the interface unit operable to:
obtain an initial state associated with the synthetic fuel generation plant, the initial state comprising state information indicative of a state of one or more processes of the synthetic fuel generation plant at an initial time point;
a data contextualisation unit configured to obtain contextual information regarding the initial state from a knowledge base and augment the data regarding the initial state with the contextual information;
a digital twin representative of the synthetic fuel generation plant, the digital twin comprising a trained machine learning model configured to:
receive a first state associated with the synthetic fuel generation plant at a first time point and the associated contextual information; and
predict a first future state, wherein the first future state comprises predicted state information indicative of a future state of one or more processes of the synthetic fuel generation plant at a second time point subsequent the first time point; an optimisation unit in communication with the interface unit and the digital twin, wherein the optimisation unit is configured to:
obtain an objective function which in use maps from a given state to an objective value for said given state, wherein the objective value for the given state is indicative of an estimated quantity and/or quality of synthetic fuel generated by the one or more processes when the synthetic fuel generation plant is in the given state; and
perform an optimisation process to determine an updated state from the initial state, the optimisation process operable to:
determine an initial future state for the initial state using the digital twin, wherein the initial future state comprises predicted state information indicative of the future state of one or more processes of the synthetic fuel generation plant at a future time point subsequent the initial time point; and
optimise the objective function to obtain the updated state such that a first objective value determined by the objective function for an updated future state is greater than a second objective value determined by the objective function for the initial future state;
wherein the updated future state is determined from the updated state by the digital twin, and the updated future state comprises predicted state information indicative of the future state of one or more processes of the synthetic fuel generation plant at the future time point; and
a reasoning support unit configured to provide an evaluation of the updated future state based on the updated future state information and the contextual information.

### Claim 2 (depends on 1)

The control system of claim 1 further comprising:
an action unit configured to:
cause the synthetic fuel generation plant to transition to the updated state determined by the optimisation process.

### Claim 3 (depends on 2)

The control system of claim 2 wherein the action unit is further configured to: determine an action plan to transition the synthetic fuel generation plant from the initial state to the updated state.

### Claim 4 (depends on 3)

The control system of claim 3 wherein, in order to cause the synthetic fuel generation plant to transition to the updated state, the interface unit is further configured to: provide the action plan to the synthetic fuel generation plant for execution by the synthetic fuel generation plant.

### Claim 5 (depends on 3)

The control system of claim 3 wherein the action plan comprises a command to control an actuator of the synthetic fuel generation plant, wherein the actuator is associated with a processes of the synthetic fuel generation plant.

### Claim 6 (depends on 1)

The control system of claim 1 wherein the state information comprises a sensor value associated with a sensor of the synthetic fuel generation plant, wherein the sensor is associated with a process of the synthetic fuel generation plant.

### Claim 7 (depends on 1)

The control system of claim 1 wherein the trained machine learning model is a supervised machine learning model trained on a dataset of historical states, wherein the dataset of historical states is obtained from a plurality of fuel generation plants.

### Claim 8

(canceled)

### Claim 9 (depends on 1)

The control system of claim 1, wherein the action unit is further configured to:
cause a remote synthetic fuel generation plant to transition to the updated state determined by the optimisation process.

### Claim 10 (depends on 9)

The control system of claim 9 wherein the action unit is further configured to:
determine a remote action plan to transition the remote synthetic fuel generation plant to the updated state.

### Claim 11 (depends on 10)

The control system of claim 10 wherein:
the interface unit is further configured to be communicably coupled to the remote synthetic fuel generation plant; and
in order to cause the remote synthetic fuel generation plant to transition to the updated state the interface unit is further configured to:
provide the remote action plan to the remote synthetic fuel generation plant for execution by the remote synthetic fuel generation plant.

### Claim 12 (depends on 1)

The control system of claim 1 wherein optimisation of the objective function is constrained by a predefined constraint.

### Claim 13 (depends on 12)

The control system of claim 12 wherein the predefined constraint in use restricts possible values of the state information of the updated state.

### Claim 14 (depends on 1)

The control system of claim 1 wherein the reasoning support system further comprises a natural language processor configured to provide a natural language output based on the updated future state information and the contextual information.

### Claim 15 (depends on 1)

The control system of claim 1 wherein the data contextualisation unit is further configured to define an ontology for the initial state defined from said knowledge base and supplements the initial state with the ontology.

### Claim 16 (depends on 15)

The control system of claim 15 wherein the data contextualisation unit is further configured to update state information to the ontology.

### Claim 17 (depends on 1)

The control system of claim 1 wherein the system further comprises a data transformation unit configured to remove anomalies from data, and wherein the data transformation unit is configured to remove anomalies from the data by statistical analysis.

### Claim 18

(canceled)

### Claim 19 (depends on 17)

The control system of claim 17 wherein the data transformation unit is further configured to combine multiple data sources.

### Claim 20 (depends on 17)

The control system of claim 17 wherein the data transformation unit is further configured to identify contextual information regarding the data from pre-existing knowledge bases and add the contextual information to the data.

### Claim 21

A computer-implemented method for optimising a fuel generation process performed by a synthetic fuel generation plant, the computer-implemented method comprising:
obtaining an initial state associated with the synthetic fuel generation plant, the initial state comprising state information indicative of a state of one or more processes of the synthetic fuel generation plant at an initial time point;
determining contextual information regarding the initial state from a knowledge base and augment the data regarding the initial state with the contextual information;
obtaining a digital twin representative of the synthetic fuel generation plant, the digital twin comprising a trained machine learning model which in use:
receives a first state associated with the synthetic fuel generation plant at a first time point; and
predicts a first future state, wherein the first future state comprises predicted state information indicative of a future state of one or more processes of the synthetic fuel generation plant at a second time point subsequent the first time point;
obtaining an objective function which in use maps from a given state to an objective value for said given state, wherein the objective value for the given state is indicative of an estimated quantity and/or quality of synthetic fuel generated by the one or more processes when the synthetic fuel generation plant is in the given state;
performing an optimisation process to determine an updated state from the initial state, the optimisation process comprising:
determining an initial future state for the initial state using the digital twin, wherein the initial future state comprises predicted state information indicative of the future state of one or more processes of the synthetic fuel generation plant at a future time point subsequent the initial time point; and
optimising the objective function to obtain the updated state such that a first objective value determined by the objective function for an updated future state is greater than a second objective value determined by the objective function for the initial future state;
wherein the updated future state is determined from the updated state by the digital twin, and the updated future state comprises predicted state information indicative of the future state of one or more processes of the synthetic fuel generation plant at the future time point; and
providing an evaluation of the updated future state based on the updated future state information and the contextual information.

### Claim at position 22

-38. (canceled)

### Claim 39

A non-transitory computer readable storage medium comprising one or more program instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of:
obtaining an initial state associated with a synthetic fuel generation plant, the initial state comprising state information indicative of a state of one or more processes of the synthetic fuel generation plant at an initial time point;
determining contextual information regarding the initial state from a knowledge base and augment the data regarding the initial state with the contextual information;
obtaining a digital twin representative of the synthetic fuel generation plant, the digital twin comprising a trained machine learning model which in use:
receives a first state associated with the synthetic fuel generation plant at a first time point; and
predicts a first future state, wherein the first future state comprises predicted state information indicative of a future state of one or more processes of the synthetic fuel generation plant at a second time point subsequent the first time point;
obtaining an objective function which in use maps from a given state to an objective value for said given state, wherein the objective value for the given state is indicative of an estimated quantity and/or quality of synthetic fuel generated by the one or more processes when the synthetic fuel generation plant is in the given state;
performing an optimisation process to determine an updated state from the initial state, the optimisation process comprising:
determining an initial future state for the initial state using the digital twin, wherein the initial future state comprises predicted state information indicative of the future state of one or more processes of the synthetic fuel generation plant at a future time point subsequent the initial time point; and
optimising the objective function to obtain the updated state such that a first objective value determined by the objective function for an updated future state is greater than a second objective value determined by the objective function for the initial future state;
wherein the updated future state is determined from the updated state by the digital twin, and the updated future state comprises predicted state information indicative of the future state of one or more processes of the synthetic fuel generation plant at the future time point; and
providing an evaluation of the updated future state based on the updated future state information and the contextual information.
